Climate change is making natural hazards more intense and more likely to occur together, but most current risk assessments still look at these hazards one at a time. This study introduces a machine learning method that can detect and track overlapping hazard events, such as floods, storms, or heatwaves, that may lead to future disasters. The approach was tested using historical climate data and successfully identified the July 2021 floods in Europe. The model was then applied to future climate projections under three scenarios, analyzing how hazards might cluster in space and time across Europe. We focused on three types of hazard combinations: (1) flood conditions caused by heavy rain and saturated soils, (2) coastal storms involving wind, rain, and low pressure, and (3) extreme heat and drought with potential for wildfires. Early findings show that under high-emissions scenarios, extreme rainfall events could become 15% more frequent and cover 18% more land, with dangerous weather shifting northward. Although results depend on how thresholds and clustering settings are chosen, further testing will refine the method. This framework provides a powerful, scalable way to track how combined climate hazards are evolving, helping planners and policymakers make better decisions in a changing world.